Some Suggestions To Eliminate Peg Maintenance Expenditures

[Ben]

I like the direction that this concept is moving in. I’m not that well versed in the underlying economics to be able to give it a 100% approval, but I’ll keep following along with this discussion and support the community with my votes if it is summarized in a motion that details the action steps required to make this a reality.

Thank you, @Benjamin, for introducing it and to the rest of the community for the enlightening discussion that has followed.

[masterOfDisaster]

This is in fact a cross post, but the important part of it is related exactly to this discussion regarding new concepts (on protocol!) for peg maintenance.
I beg your forgiveness for cross posting.
I hope the post supports the discussion though.

Park Rate Voting

This is why Nu needs a feedback loop on protocol level and not only parking interest and motions based on what NSR holders think might be appropriate. At the moment there's no immediate and ongoing feedback between decisions (vote for 0% park rate; vote for x NSR to be sold to buy NBT and burn them). An effect is that Nu faces parking rates that are higher than necessary and NSR sales that with bigger amounts than necessary to stabilize the peg. Once the B&C Exchange funding is through Nu …

[Nagalim]

Why don’t we seed both sides every day? It doesn’t give the same kind of manual leverage over the peg, but it does allow easy flow between nbt and nsr. We could have 2 bits in the blockchain instead of 1: both seeds are on by default and a 50% or greater vote can turn off either seed or both.

[mhps]

Any funds sent to the correct NSR or NBT address is credited (just like with NuLagoon). ‘Burns’ are sent back to reserve.

Such resereve-based centralized operations essentially is just a shareholder funded LPC on the NBT/NSR pair, if the reserve fund is printed. Such operations had all sorted of problems and was closed off.

If the reserve is produced by pooling members’ funds together, then isn’t it just a NuLagoon on NBT/NSR with 6,000 NBT on the sell side and ~3 million NSR on the buy side? It is unlikely to make much difference than current peg operations (30k - 40k NBT on each side).

Open ended burning is what is needed.

[Benjamin]

Why don’t we seed both sides every day? It doesn’t give the same kind of manual leverage over the peg, but it does allow easy flow between nbt and nsr. We could have 2 bits in the blockchain instead of 1: both seeds are on by default and a 50% or greater vote can turn off either seed or both.

I strongly prefer seeding only one asset type for the following reasons:

1. You are adding an additional alternative. Asking people to vote over three alternatives may result in the group selecting an undersirable outcome. [http://en.wikipedia.org/wiki/Arrow’s_impossibility_theorem][1]
   [1]: http://arrow%20impossibility%20theorem%20wikipedia

Specifically, even if every single voter agrees that outcome A is better than outcome B, the group can still end up voting for B.

2. Seeding a single currency is effectively creating a subsidy for private liquidity providers to enter the market. If you seed both currencies, then there is no longer any incentive for private participation. This would hinder the development of a liquid NSR market.

[Benjamin]

Such resereve-based centralized operations essentially is just a shareholder funded LPC on the NBT/NSR pair. Such operations had all sorted of problems and was closed off.

This in incorrect. The LPC system directly controls the market price of the NBT/NSR pair. In the auction system, supply is controlled and the price is determined by market equilibrium. The auction system incentivizes private market making activity, i.e. profits from arbitrage go to currency speculators. The LPC system controls market making activity directly, profits from arbitrage are ruled out by price controls.

[mhps]

The auction system incentivizes private market making activity, i.e. profits from arbitrage go to currency speculators. The LPC system controls market making activity directly, profits from arbitrage are ruled out by price controls.

Thanks for pointing it out. I like the direction of the proposal made in the OP but altering the protocol is risky and no small task so it would be best if we could bootstrap the process by demonstrating.

[Nagalim]

The biggest differences sending to reserve rather than burning has are:
A) centralization
B) one person picks nsr or nbt rather than the blockchain
C) a grant will be needed when reserves on one side run low
D) auction volume will not have a feedback loop (this is somewhat trivial in my opinion)
E) we are actually already prepared to implement it if we use a reserve, we don’t even have to wait for nsr grants.

We could implement this thing tomorrow without too much difficulty, honestly.

[Nagalim]

Could we institute this with a small seed and slowly ramp up? Like start with 50NBT/25000NSR and increase by 50NBT per day until 1% (5 months). This would allow a price to be built over a little time and get word out. If the Auctioneer announces the seed 4 hours before auction result, it will be a good test and hopefully get people used to it (or there could be a continual announcement that cannot be flipped within 4 hours of the auction closing). Again, bidding for the auction is as simple as sending funds to NSR and NBT addresses. If you send a proportional amount of NSR and NBT to equal your ideal price, you will always win from your own perspective.

By the way, my above example was totally wrong. If Lucy is the only one participating in the auction she wins all 10 NSR just by bidding .01 NBT.

Anyway, I just really want to see this be a motion. The technical challenges of being an auctioneer are just having NBT and NSR addresses, having access to the reserves, and doing a manysend that is proportional to the recieved outputs for the day.

[Nagalim]

Is this alright? If not, please let me know, I’d be happy to step down for someone else to hash and word the motion. I’m just really scared that we’re going to continue with the mandatory NSR selloff.

[Draft] Seeded Auctions

Custodial hash RIPEMD160 hash: tbd =##=##=##=##=##=## Motion hash starts with this line ##=##=##=##=##=##= 1350 nbt @Nagalim will continue running automated auctions every 5000 blocks and will seed the auctions with 200 nbt per auction for 25000 blocks. Received nsr will be burnt. @Nagalim will be rewarded 200 nbt for open sourcing the software, 100 nbt for acting as auctioneer, and 50 nbt for acting as reserve. =##=##=##=##=##=## Motion hash ends with this line ##=##=##=##=##=##= Tria…

[mhps]

By the way, my above example was totally wrong. If Lucy is the only one participating in the auction she wins all 10 NSR just by bidding .01 NBT.

Would you mind modifying that post to show a corrected and easy to understand version of how it works? I think i understand what you meant to do but I am not sure as I have a hard time parsing something like

But this will only keep her neutral with the nbt demurrage, so she goes a tiny bit more to even the peg gradually over time.

[Nagalim]

I edited that post.

“The result of this is that Nu gets to sell its NSR directly to Lucy for relevant prices without having to trust Lucy at all.”

If we go further and take Lucy out of the picture, we will wind up with @Benjamin’s concept, that NBT will deviate from $1 and that deviation will use the auction as a lever to right the peg. One of the key points I’m trying to make is that we don’t need the deviation from $1 if we have liquidity targets with compensation because deviation from a balanced peg acts as a virtual deviation from $1 financially.

[mhps]

I edited that post.

Thanks. The post is much easier to understand now.

[Nagalim]

Alright, I am risking the further confusion of my potential participant base here, but I think it’s important we discuss possible alternative equations for the auctions.

Currently, the procedure is to turn all received nbt into nsr and all received nsr into nbt at the auction price, then distribute.

A possible alternative is actually taken from something @mhps suggested back before the first unseeded auction. The idea would be to treat the auction as a basket of funds and to take a percentage of it to distribute to each participant. For example, if A submits 10 NSR and B submits 10 NBT, A has submitted 100% of the nsr so therefore 50% of the funds. Therefore, A is awarded 5 nbt and 5 nsr. B is awarded the same for submitting 100% of the nbt.

In the current auctions, the above scenario would result in A getting 10 nbt and B getting 10 nsr. I’m still processing the consequences and potential attack vectors, but would appreciate more minds thinking about it.

Edit: I gave this some thought and have come to the conclusion that this method is equivalent to the currently implemented auctions, except that it is exactly half as efficient of a conversion process. Therefore the current implementation is superior to this alternative.

[mhps]

Edit: I gave this some thought and have come to the conclusion that this method is equivalent to the currently implemented auctions, except that it is exactly half as efficient of a conversion process. Therefore the current implementation is superior to this alternative.

You can’t convince anyone if you don’t give a step by step analysis that can be understood and verified.

[Nagalim]

The full generalized math for this auction goes like this:

1. We have an auction price before submission P
2. We have an auction volume before submission V
3. We have a submission price Q
4. We have a submission volume W
5. We have an auction price after submission P’
6. P’ = (PV + QW) / (V+W)
7. We have a conversion volume R relating to the effectiveness of the auction
8. In the current method: Rc = W (1 - Q / P’)
9. In the new method: Rn = W - (W+V)*[W / (W+V) + (Q/P’) * W / (W+V)] / 2 = W (1 - (1 +Q / P’) / 2)
10. Therefore, the new method simply averages the relative price (submission/auction) with unity. Its purpose and effect would be to desensitize the auction results to variance among the submission prices.
11. For unseeded auctions, this normalizing effect may actually be beneficial as participation is extremely low. For seeded auctions, however, it ends up with Nu subsidizing auction participation and is generally bad financial practice. This can be seen in that if the new method were implemented every auction Nu would submit only one currency and would receive both, proving that there must be some wasted efficiency.
12. We can make the normalizing effect arbitrarily powerful (add 2 and divide by 3, add 100 and divide by 101, etc.) until the point where all participants simply receive back any funds they send.

[mhps]

Are you referring to this post of mine?

We have a conversion volume R relating to the effectiveness of the auction

can you define R in words so your math can be verified against the definition?

[Nagalim]

Yes that is the post I am referring to.
R = number of nbt sent - number of nbt received. This number will be positive for converting nbt into nsr (Q < P’).

[mhps]

R = number of nbt sent - number of nbt received.

Does “Rc = W (1 - Q / P’)” mean that the submission volume W only has nbt? Where are the nsr in a general case?

[Nagalim]

Let’s say W = 10 NBT, Q = $1, P’ = $2. That means R = 5 NBT. Since we know the prices, we know that 10 NSR were submitted and 20 NSR were received, giving the NSR version of R as -10 NSR. You can also get this by taking R and multiplying by -P’. The NSR version of R is just how much NSR the participant received for paying R NBT.

Stating R and P’ is sufficient to know how much NBT was sold at what price, which implies the cost.